On Mon, Mar 03, 2025 at 07:22:17AM +0000, Gavin Smith wrote: > On Sun, Mar 02, 2025 at 10:46:26PM +0100, Patrice Dumas wrote: > > Hello, > > > > The format of the test in tta/perl/t/*.t is going to change, replacing > > Perl tree structure with a textual representation of the tree, which > > should contain the same information. The corresponding commit will be a > > big commit, and we cannot really know if there is gonna be some errors > > in the new trees as nobody will check that the old and new representation > > match for all the tests, but everything I had a look at looked good. > > Are you generating the new representation from the old representation > using a Perl program, or generating the new representation anew by > running the tests?
The second one. But in a way it is also the first one, as the old presentation was the tree as Perl structures, and what the new code do is to give it a textual representation, so it would probably have been the same code used to do the first one, only called differently. > If the former there would be less chance of an > unnoticed change. But as the main program itself is not changing, I > think there is little chance of a bug being introduced either way. Agreed. The risk is not on bugs introduced, but on errors in the representation that would lead to a regression not being discovered later on. I am pretty sure that it is not the case, but I may have missed something. -- Pat
